fix(no-barrel-import): framework-aware wording for React Native targets

[devin-ai-integration]

Summary

no-barrel-import's message says the barrel import "slows page load" — meaningless for React Native, which has no page. The message lead sentence now branches on the file's platform target:

• RN/Expo target → This ships extra code in your app bundle & slows startup.
• web (or unknown) target → existing This ships extra code to your users & slows page load.

The branch condition is the testable part: it reuses the exact decision ladder behind isReactNativeFileActive (native .ios/.android/.native extensions → nearest package.json platform → project framework setting), now exposed as a three-way classifier so wording can default to web on "unknown" while RN rule gating keeps its conservative "unknown → active" behavior:

classifyReactNativeFileTarget(context): "react-native" | "web" | "unknown"
isReactNativeFileActive = (context) => classifyReactNativeFileTarget(context) !== "web"

Notably, .web.tsx files inside an RN monorepo still get the page-load wording, and projects with no discoverable framework keep the existing wording (no behavior change for any existing web diagnostics — the source-resolution fallback message is framework-neutral and unchanged).

Unit tests cover each branch: nextjs → page-load, react-native/expo → app-startup, unknown → page-load, and App.web.tsx under framework: "react-native" → page-load.

The "is this wording better" editorial half (e.g. the React 19 caveat phrasing) is intentionally left out of this PR.

Link to Devin session: https://app.devin.ai/sessions/5ebb159406b24bf0bd4f4d515b3cfbf8
Requested by: @aidenybai

---

Note

Low Risk
User-facing lint message text only; RN rule activation behavior is preserved via isReactNativeFileActive !== "web".

Overview
no-barrel-import diagnostic copy now depends on the file’s platform target instead of always mentioning page load. React Native / Expo targets get “ships extra code in your app bundle & slows startup”; web, unknown framework, and .web.* files in RN monorepos keep the existing “slows page load” wording.

Platform detection is centralized by refactoring is-react-native-file: new classifyReactNativeFileTarget returns "react-native" | "web" | "unknown" (same ladder as before: native/web extensions → nearest package.json → framework setting). isReactNativeFileActive is now !== "web" so RN rule gating stays conservative on "unknown" while messaging treats "unknown" as web.

Adds no-barrel-import.test.ts covering nextjs, react-native, expo, unknown, and App.web.tsx under RN; patch changeset for oxlint-plugin-react-doctor.

^{Reviewed by Cursor Bugbot for commit cdd477b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/eslint-plugin-react-doctor@796

npm i https://pkg.pr.new/oxlint-plugin-react-doctor@796

npm i https://pkg.pr.new/react-doctor@796

commit: cdd477b

[github-actions]

No React Doctor issues found. 🎉

_{Reviewed by React Doctor for commit cdd477b.}

[devin-ai-integration]

Ran end-to-end CLI testing of the framework-aware wording: built the CLI from this branch and ran react-doctor --json --no-score --no-dead-code against two minimal fixture projects with identical barrel-import source, differing only in package.json deps.

• RN project (react-native dep) → app-startup wording — passed
• Web project (react-dom/vite deps) → existing page-load wording — passed (regression)
• src/Other.web.tsx inside the RN project → page-load wording while App.tsx keeps app-startup wording in the same run — passed

RN project output (the change)

{ "f": "src/App.tsx",
  "m": "This ships extra code in your app bundle & slows startup. Import directly from \"./components/Button\"." }

Web project output (regression — unchanged)

{ "f": "src/App.tsx",
  "m": "This ships extra code to your users & slows page load. Import directly from \"./components/Button\"." }

Mixed run: .web.tsx inside RN project

[
  { "f": "src/App.tsx",       "m": "This ships extra code in your app bundle & slows startup. Import directly from \"./components/Button\"." },
  { "f": "src/Other.web.tsx", "m": "This ships extra code to your users & slows page load. Import directly from \"./components/Button\"." }
]

Testing was CLI/shell-only, so evidence is JSON output rather than screenshots. Devin session